feat: 金额 decimal 精确计费、failover 按失败类型封顶、渠道小时桶预聚合、分层守护测试
- 计费：19 个金额字段 float64→decimal.Decimal，DB 列保持 numeric(20,8) 零迁移； API 金额仍为 JSON number（MarshalJSONWithoutQuotes），前端零改动 - failover：保留时间预算，新增 OT_PROXY_FAILOVER_CAPS_NET/SERVER/RATE/QUOTA 按失败类型封顶换渠道次数（429/配额类瞬时失败时间预算形同虚设） - 预聚合：新增 usage_channel_hourlies（schema v10），记账事务内同步 upsert， 长窗口渠道统计走桶不再扫明细，spent_last_30d 改扫 usage_daily，v10 一次性回填幂等 - 守护：internal/arch_test.go 用 go/parser 强制分层依赖（fail-closed）
This commit is contained in:
Sakurasan committed
1 parent
5d6e5bdb04
commit
b2e500ff51
25 files changed
+841
-253
No files matched your search
@@ -10,6 +10,7 @@ import (
|
||||
"github.com/gin-gonic/gin"
|
||||
"github.com/openteam/server/internal/pkg/resp"
|
||||
"github.com/openteam/server/internal/store"
|
||||
"github.com/shopspring/decimal"
|
||||
)
|
||||
|
||||
// AdminStatsOverview GET /api/v1/admin/stats/overview — 运营总览。
|
||||
@@ -29,7 +30,7 @@ func (h *Handler) AdminStatsOverview(c *gin.Context) {
|
||||
|
||||
// 全局今日/本月汇总（跨用户）
|
||||
var todayReq, monthReq int64
|
||||
var todayCost, monthCost float64
|
||||
var todayCost, monthCost decimal.Decimal
|
||||
var todayTokens, monthTokens int64
|
||||
h.a.DB.Model(&store.UsageDaily{}).Where("date = ?", today).
|
||||
Select("COALESCE(SUM(requests),0), COALESCE(SUM(cost),0), COALESCE(SUM(input_tokens+output_tokens+cache_read_tokens),0)").
|
||||
@@ -42,7 +43,7 @@ func (h *Handler) AdminStatsOverview(c *gin.Context) {
|
||||
var days []struct {
|
||||
Date string
|
||||
Req int64
|
||||
Cost float64
|
||||
Cost decimal.Decimal
|
||||
}
|
||||
h.a.DB.Model(&store.UsageDaily{}).
|
||||
Select("date, SUM(requests) req, SUM(cost) cost").
|
||||
@@ -64,6 +65,10 @@ func (h *Handler) AdminStatsOverview(c *gin.Context) {
|
||||
|
||||
// AdminChannelStats GET /api/v1/admin/channels/stats?window=1h — 按渠道聚合近一段的
|
||||
// 请求量、成功率、错误码分布与延迟分位。定位"哪条渠道在掉"不该靠翻日志。
|
||||
//
|
||||
// 短窗口（5m/1h）扫 usage_logs 保留精确延迟分位与错误码分布；长窗口（6h/24h/7d）
|
||||
// 走 usage_channel_hourly 小时桶——明细扫 5 万行封顶在 7d 下只剩个尾巴，桶是全量，
|
||||
// 且不受明细保留期清理影响。长窗口延迟给 avg/max（p50/p95 以 avg 填充，桶里没有分布）。
|
||||
func (h *Handler) AdminChannelStats(c *gin.Context) {
|
||||
win := c.DefaultQuery("window", "1h")
|
||||
d, ok := statWindows[win]
|
||||
@@ -72,40 +77,11 @@ func (h *Handler) AdminChannelStats(c *gin.Context) {
|
||||
}
|
||||
since := time.Now().UTC().Add(-d)
|
||||
|
||||
// 只取窗口内最近这么多行，避免大窗口把内存吃光（超出部分不参与分位计算）
|
||||
const maxRows = 50000
|
||||
var rows []channelStatRow
|
||||
h.a.DB.Model(&store.UsageLog{}).
|
||||
Select("channel_id, status, latency_ms, cost, "+
|
||||
"(input_tokens + output_tokens + cache_read_tokens) AS tokens, error_code, created_at").
|
||||
Where("created_at >= ? AND channel_id > 0", since).
|
||||
Order("id DESC").Limit(maxRows).Scan(&rows)
|
||||
|
||||
agg := map[uint64]*chanAgg{}
|
||||
for _, r := range rows {
|
||||
a := agg[r.ChannelID]
|
||||
if a == nil {
|
||||
a = &chanAgg{lat: make([]int, 0, 64), errs: map[string]int{}}
|
||||
agg[r.ChannelID] = a
|
||||
}
|
||||
a.requests++
|
||||
a.cost += r.Cost
|
||||
a.tokens += r.Tokens
|
||||
switch r.Status {
|
||||
case store.UsageStatusSuccess:
|
||||
a.success++
|
||||
a.lat = append(a.lat, r.LatencyMS) // 分位只按正常完成的请求算，中断/失败的延迟不代表体感
|
||||
case store.UsageStatusCanceled:
|
||||
a.canceled++
|
||||
default:
|
||||
a.err++
|
||||
if r.ErrorCode != nil && *r.ErrorCode != "" {
|
||||
a.errs[*r.ErrorCode]++
|
||||
}
|
||||
}
|
||||
if r.CreatedAt.After(a.lastAt) {
|
||||
a.lastAt = r.CreatedAt
|
||||
}
|
||||
if d > time.Hour {
|
||||
h.aggChannelStatsHourly(since, agg)
|
||||
} else {
|
||||
h.aggChannelStatsRaw(since, agg)
|
||||
}
|
||||
|
||||
var chans []store.Channel
|
||||
@@ -137,6 +113,75 @@ func (h *Handler) AdminChannelStats(c *gin.Context) {
|
||||
resp.OK(c, gin.H{"window": win, "since": since, "items": out})
|
||||
}
|
||||
|
||||
// aggChannelStatsRaw 短窗口：扫明细（限最近 5 万行防内存失控），保留精确分位与错误码。
|
||||
func (h *Handler) aggChannelStatsRaw(since time.Time, agg map[uint64]*chanAgg) {
|
||||
const maxRows = 50000
|
||||
var rows []channelStatRow
|
||||
h.a.DB.Model(&store.UsageLog{}).
|
||||
Select("channel_id, status, latency_ms, cost, "+
|
||||
"(input_tokens + output_tokens + cache_read_tokens) AS tokens, error_code, created_at").
|
||||
Where("created_at >= ? AND channel_id > 0", since).
|
||||
Order("id DESC").Limit(maxRows).Scan(&rows)
|
||||
|
||||
for _, r := range rows {
|
||||
a := agg[r.ChannelID]
|
||||
if a == nil {
|
||||
a = &chanAgg{lat: make([]int, 0, 64), errs: map[string]int{}}
|
||||
agg[r.ChannelID] = a
|
||||
}
|
||||
a.requests++
|
||||
a.cost = a.cost.Add(r.Cost)
|
||||
a.tokens += r.Tokens
|
||||
switch r.Status {
|
||||
case store.UsageStatusSuccess:
|
||||
a.success++
|
||||
a.lat = append(a.lat, r.LatencyMS) // 分位只按正常完成的请求算，中断/失败的延迟不代表体感
|
||||
case store.UsageStatusCanceled:
|
||||
a.canceled++
|
||||
default:
|
||||
a.err++
|
||||
if r.ErrorCode != nil && *r.ErrorCode != "" {
|
||||
a.errs[*r.ErrorCode]++
|
||||
}
|
||||
}
|
||||
if r.CreatedAt.After(a.lastAt) {
|
||||
a.lastAt = r.CreatedAt
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// aggChannelStatsHourly 长窗口：读渠道×小时桶。p50/p95 用 avg 近似（桶里没有延迟分布），
|
||||
// 错误码分布桶里没有、置空；last_at 取最新桶的起始小时（精度 1h）。
|
||||
func (h *Handler) aggChannelStatsHourly(since time.Time, agg map[uint64]*chanAgg) {
|
||||
var rows []store.UsageChannelHourly
|
||||
// 桶粒度 1h：包含 since 落在的那个小时（边界桶会带上窗口前的一小段，可接受）
|
||||
h.a.DB.Where("hour >= ?", since.UTC().Format("2006-01-02T15")).Find(&rows)
|
||||
|
||||
for _, r := range rows {
|
||||
a := agg[r.ChannelID]
|
||||
if a == nil {
|
||||
a = &chanAgg{lat: make([]int, 0, 64), errs: map[string]int{}}
|
||||
agg[r.ChannelID] = a
|
||||
}
|
||||
a.requests += r.Requests
|
||||
a.success += r.Success
|
||||
a.canceled += r.Canceled
|
||||
a.err += r.Errors
|
||||
a.cost = a.cost.Add(r.Cost)
|
||||
a.tokens += r.InputTokens + r.OutputTokens + r.CacheReadTokens
|
||||
if r.Success > 0 {
|
||||
avg := int(r.LatencySum / r.Success)
|
||||
a.lat = append(a.lat, avg)
|
||||
}
|
||||
if r.LatencyMax > 0 {
|
||||
a.lat = append(a.lat, r.LatencyMax) // 参与分位计算，max 取两者较大值
|
||||
}
|
||||
if bucketAt, err := time.Parse("2006-01-02T15", r.Hour); err == nil && bucketAt.After(a.lastAt) {
|
||||
a.lastAt = bucketAt
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// statWindows 支持的统计窗口。
|
||||
var statWindows = map[string]time.Duration{
|
||||
"5m": 5 * time.Minute,
|
||||
@@ -150,7 +195,7 @@ type channelStatRow struct {
|
||||
ChannelID uint64
|
||||
Status string
|
||||
LatencyMS int
|
||||
Cost float64
|
||||
Cost decimal.Decimal
|
||||
Tokens int64
|
||||
ErrorCode *string
|
||||
CreatedAt time.Time
|
||||
@@ -158,7 +203,7 @@ type channelStatRow struct {
|
||||
|
||||
type chanAgg struct {
|
||||
requests, success, err, canceled int64
|
||||
cost float64
|
||||
cost decimal.Decimal
|
||||
tokens int64
|
||||
lat []int
|
||||
errs map[string]int
|
||||
|
||||
Reference in new issue
Block a user